Feed

Cell-cell interactions influence gene expression in the tumor microenvironment, yet despite advances in spatial transcriptomics that enable in situ expression profiling, current computational tools remain limited in quantitatively resolving single-cell and multicellular interaction effects across patient tissues. Here, we present InSituPREP (In Situ Proximity Expression Programs), a computational framework for quantifying how spatial context relates to transcriptional states in three-dimensional tissues at single-cell resolution. The framework quantifies proximity-associated gene programs across multiple analytical layers. We applied InSituPREP to Expansion Sequencing data from breast cancer biopsies of 10 patients, integrating newly generated and previously published datasets and profiling 299 cancer-related genes. This analysis revealed that a portion of within-cell type variability reflects proximity-associated transcriptional states, that proximity-dependent programs are reproducible across patients and measurement platforms, and that spatial RNA velocity links intercellular distance to future transcriptional trajectories. Linear modeling showed that gene expression in specific cell types varies with intercellular distance and the number of neighboring cells, while triplet configurations revealed additional interaction-associated programs. Moreover, simultaneous detection of bacterial signals and host mRNA enabled us to link bacterial proximity to distinct transcriptional responses. InSituPREP provides a framework for quantifying spatially associated gene expression and dissecting context-dependent transcriptional states at single-cell resolution.